I have epiphany built from git master today. There is an odd cursor appearing
when mouse pointer is on an image that is opened inside epiphany. If I take a
screenshot, in the screenshot the normal mouse pointer appears.
Epiphany maintainer confirms that this is reproducible on GtkLauncher and MiniBrowser
Both the zoom-in and the zoom-out cursors seem to be corrupted somehow. I haven't checked the other custom cursos, though.
After the patch in bug 118905 landed there is no "ghost" cursor anymore, but no cursor either.
Created attachment 210725 [details]
I wrote this small test case from the code in CursorGtk.cpp, it shows that the cursor is always corrupted. Something has changed that is rendering the code useless.
So it was a GTK+ bug. I fixed it in GTK+ master already. However there is a method we should be using here to make this simpler. gdk_cursor_new_from_surface(). I'll cook a patch.
Is gdk_cursor_new_from_surface in GDK 3.6?
No, only 3.9. But I think we want to use it for hires display support.
If we're not going to bump to 3.9 for this, then I guess we can close this already.
(In reply to comment #7)
> If we're not going to bump to 3.9 for this, then I guess we can close this already.
Why? We can use the new code when GTK version is >= 3.9, no?
(In reply to comment #8)
> (In reply to comment #7)
> > If we're not going to bump to 3.9 for this, then I guess we can close this already.
> Why? We can use the new code when GTK version is >= 3.9, no?
If we are interested in keeping backward compatibility, we can just write a version that targets <= 3.6 and use it conditionally. It's just a question of whether or not we think it's useful / makes the code nicer.
Created attachment 211011 [details]
Comment on attachment 211011 [details]
Clearing flags on attachment: 211011
Committed r155329: <http://trac.webkit.org/changeset/155329>
All reviewed patches have been landed. Closing bug.